Survey of Physician Attitudes on Artificial Intelligence

Objective To better understand how physicians view the potential use of artificial intelligence (AI) in their practices. Method We conducted an online cross-sectional survey of 1,258 physicians (MD, DO, MBBS or equivalent) in the United States (US) and Canada from December 2023 through February 2026. The survey contained 34 questions focused on 6 areas: (1) demographics, (2) training, (3) clinical benefits, (4) clinical concerns, (5) clinical AI prediction tools, and (6) generative AI. Results A total of 130 completed surveys were received (response rate, 10%). Top specialties included psychiatry (46.2%), internal medicine (14.6%) and pediatrics (13.8%). A total of 16.9% had training in computer science or information technology and 40.8% in statistics. Nearly half (46.9%) had no formal training in AI and were self-taught. Those with training in AI received it through vendors or at work. Sixty to seventy percent thought AI would provide diagnostic aid or treatment recommendations, but half did not feel adequately trained to safely use AI in practice. Over 70% thought it could help with administrative tasks. More than 80% were worried about errors, misinformation, or biases in the data. Conclusion There is a growing sense that AI in medicine could be helpful with specific tasks. Ongoing training, education, and study are needed to determine how AI will best integrate into clinical practice.